| ??? 04/13/09 20:29 Read: times Msg Score: +1 +1 Informative |
#164557 - I still don't know what you meant, Andy ... Responding to: ???'s previous message |
Andy Neil said:
Richard Erlacher said:
one can easily run the tools under LINUX, which is an adequate native-mode environment for ARM. Unfortunately, it will always drive the cost of a subsystem based on the MCU upward in cost and complexity. Andy Neil said:
So don't do it!
Just because it can be done doesn't mean that it has to be done! Richard Erlacher said:
I'm not following what you're saying ... could you be more specific? I thought you were saying that the problem with using the target also as your development platform was that doing so increased the target cost - so the answer to that is, surely, just to not use the target also as your development platform! It's not like PCs are scarce or hard to obtain... Using a development board as your target is not what I meant at all! However, PC's seldom can be forced to "Stay out of the way." I prefer not to rely on software that runs under Windows. I can't always avoid it, but ... and PC-LINUX isn't totally better. There's an indefinite antecedent in your comment. You say "Don't do it" but don't define what you mean by it. I have, on numerous occasions, used a development platform that was a vast superset of the target environment as a development tool. The key is to have an adapter from the development platform bus to the target board. I've generally found it pretty easy to generate such a thing. The advantage is that, given the system clock rates are compatible, which they frequently are, one then can essentially exercise one's code in the target environment, under more-or-less normal circumstances. I guess it could be considered somewhat like using an ICE, albeit without the PC's t*t's in the way. Back when those processors were popular, I had such boards as would plug into nearly any target that used a Z80, 6801, 6502 (more difficult due to the lack of a tristate bus), 6802, 6809, 68008, 8086, etc. The Maxim/Dallas DS89C4x0 drop-in parts allow similar modes of operation for 805x, so long as one doesn't rely on peripherals not in that MCU. If you build a daughterboard that uses an external UART, you have a lot of ability only an expensive ICE will provide, in that you can run real time exercises on the target environment by means of a monitor that allows line-by-line disassembly and assembly in read-write memory. There are lots of things you can do, and many of them are really handy (quick-and-dirty). If you run LINUX on your ARM-based target, using such an arrangement, you can produce code that will reside in RAM and drive your target environment, provided you use at least minimal imagination to keep you from tripping over dev-board and target hardware/firmware/software features. Needless to say, this isn't for everyone. As for me, better a $200 development board than a $5k cross-compiler, especially since it doesn't simulate the target peripherals. RE |
